KR20110055636A - Messaging application with multiple viewports for presenting messages in different orders - Google Patents

Abstract

In a method for displaying messages, the system displays messages from a single user account in multiple viewports. Each viewport sorts messages through this viewport based on importance scores calculated based on the user's previous interactions with messages in his user account. Each viewport associated with a user account sorts the messages using a different message importance model.

Description

MESSAGING APPLICATION WITH MULTIPLE VIEWPORTS FOR PRESENTING MESSAGES IN DIFFERENT ORDERS}

The disclosed embodiments generally relate to the display of messages such as email, instant, and voice mail messages. More specifically, the disclosed embodiments relate to systems, methods, and computer programs for displaying messages ordered based on importance factors or scores determined from user interactions with the messages. .

As email communication increases, the number of email messages received and stored in user accounts also increases. A user account typically includes all messages sent to and from each email address or username. However, some user accounts may be associated with multiple email addresses or usernames, sometimes referred to as aliases, which together may be considered as one logical email address or username. The amount of email received can quickly overwhelm users, making it difficult to filter out important messages from unnecessary messages.

In addition, many people today access and view their email on mobile devices such as handheld computers or cell phones. Such mobile devices typically have small screens with much smaller message windows or interfaces for viewing messages. These interfaces sometimes only allow the user to view a small number of messages at any given time, requiring the user to interact with these interfaces more frequently to locate important messages, for example, by scrolling through the messages. do. These mobile devices can also take advantage of network connectivity, which is sometimes charged by usage and often slow. Users of such devices want to limit the messages they see to messages of high importance when accessing messages through such media.

To address these problems, some messages allow users to organize messages into folders or add user-defined labels for easier identification. In addition, in some email applications, users can sort messages into a particular view according to the value of a single user-defined message header field such as message delivery date, sender, or message subject. However, these organizational techniques sometimes do not identify the messages that are most important to the user and allow the user to locate the messages that are most important after scrolling through the many messages.

There is a need to provide a message system and method that addresses these shortcomings while providing the user with a customized view of messages that are automatically sorted based on the predicted importance to a particular user.

In a method for displaying messages, the system displays messages from a single user account (ie, each user's account) in multiple viewports. Each viewport sorts messages through this viewport based on importance scores calculated based on the user's previous interactions with messages in his user account. Each viewport associated with a user account sorts the messages using a different message importance model.

In some embodiments, the first client device displays messages in the viewport sorted by the first importance score calculation and the second client device displays messages from the same user account sorted by the second importance score calculation in the viewport. do. Importance prediction models are used to generate importance scores based on user interactions with messages.

In some embodiments, the client device displays messages in the viewport sorted by the first importance score calculation and messages in the second viewport from the same user account sorted by the second importance score calculation.

In another aspect of the invention, the server sends a list of messages sorted by the first importance score calculation to the first client device and messages from the same user account sorted by the second importance score calculation to the second client device. Send another list of them.

In another aspect of the invention, the server sends a list of messages sorted by the first importance score calculation to the client device and another of messages from the same user account sorted by the second importance score calculation to the client device. Send the list.

Some aspects provide a computer readable storage medium storing one or more programs comprising instructions for performing the methods.

1 is a block diagram illustrating the infrastructure of a client-server network environment in accordance with some aspects.
2 is a flow diagram illustrating a process of providing message lists arranged in multiple clients in accordance with some aspects.
3 is a flow diagram illustrating another process of providing message lists arranged into a single client in accordance with some other aspects.
4 is a flow diagram illustrating a process of generating an ordered message list in accordance with some aspects.
5A-5D are block diagrams illustrating data structures used to generate an ordered message in accordance with some aspects.
6 is a block diagram of an example client in accordance with some aspects.
7 is a block diagram of an example server in accordance with some aspects.
Like reference numerals refer to corresponding parts throughout the drawings.

1 is a block diagram of a system 100 in accordance with some aspects. One or more clients 103 and message server 106 are connected to communication network 104 via any suitable means. The communication network 104 may be one or more local area networks (LANs), wide area networks (WANs), such as the Internet, or a combination thereof.

In some aspects, client 103 is any suitable computing device such as a personal computer, handheld computer, personal digital assistant, cellular-phone, and the like. Client 103 includes one or more software applications or interfaces for viewing messages. As used herein, messages refer to any type of communication messages from one person, one station, or group to another, and include e-mail (email) messages, instant messages, voicemail messages, and the like.

In some embodiments, the message server 106 may be a front end server 120, a search engine 122, a search result ranker 124, a importance prediction model 126, a message database 128, and a user account database ( 130). Search engine 122 communicates with message database 128 to retrieve sets of messages belonging to a particular user account associated with user account database 130. The user account includes all messages sent to and from each email address or username. In some cases, each user account may be associated with multiple email addresses or usernames, sometimes referred to as aliases, which together are one logical email address or username for the purpose of this discussion. Can be considered. The search engine 122 sends a set of retrieved messages to the search result ranker 124, which search result ranker 124 calculates for each individual message in the set using the importance prediction models 126. Organize message order according to importance scores. In one case, the set of messages returned are all new messages.

Message database 128 stores messages for users. In some embodiments, one message database 128 is used per user account, and in other embodiments messages from multiple users are stored in the same message database 128. Those skilled in the art will appreciate that there are many ways to prevent messages from one user from being accessed by other users of the system.

Importance prediction models 126 calculate importance scores for individual messages in at least one user account. One or more importance prediction models 126 are associated with a user account. Importance prediction models 126 calculate importance scores for a set of one or more messages in a user account.

In one embodiment, the importance prediction models 126 are generated using machine learning. Machine learning includes a set of techniques for generating functions and prediction models, implemented using software tools and computer systems (eg, by determining weights to be applied to the components of the functions or prediction models). Machine learning is well known to those skilled in the art and is not described in this document. Each prediction model is used to calculate a importance score, or two or more importance scores, for each message in the set of messages. This model is defined to calculate one or more scores for each message using one or more message quality signals (sometimes referred to as message quality factors, or message importance factors). Each message quality signal in a set of predefined message quality signals may be weighted differently in different prediction models, where the weights applied to each message quality signal are prior user actions on the messages in the user account. This is because it depends on the user or user preferences.

Examples of message quality signals include: a signal identifying whether the user has read a message, a signal identifying a delay from the receipt of the message to the first time the user reads the message, the user receiving the message A signal identifying whether or not the user has forwarded the message, a signal identifying whether the user has forwarded the message, a signal identifying whether the user reads the message out of sequence, or the user is caught reading a specific message. A signal indicating total time, a signal identifying whether the user has performed a search for the message, a signal indicating the number of times the user has read the message, and the user's affinity for another participant of the message Signal representing (affinity). Other message quality signals that can be used to sort messages in the viewport include signals based on message header information and signals based on other metadata for the messages.

A viewport is a user interface for viewing a set of messages in a user account and interacting with the set of messages. The client may provide one or more viewports for viewing messages, that is, different viewports may exist on the same device, such as different message viewing windows on the same personal computer. The subset of available message quality signals and other message signals can be combined to produce importance scores for a particular viewport (eg, linear combination, such as weighted summation, or non-linear combination), and these importance scores are It is then used to sort the messages for the viewport.

In some embodiments, importance prediction models 126 use machine learning to establish an initial prediction model of important message characteristics for a particular viewport. This model is then applied to each message in the set of messages to calculate a importance score for that particular message. The set of messages is then sorted using search result ranker 124, ie for a particular viewport. For example, search result ranker 124 may sort the messages in order of decreasing importance score.

The choice of importance prediction models depends on the selection of a particular viewport at the client. In some embodiments, the viewport can display all the messages of the user account, as will be displayed when viewing the user's entire message inbox. In other embodiments, the viewport can display a subset of the messages in the user account, such as all messages in a particular folder or subfolder. In still other embodiments, the viewport can display a subset of messages in the user account, such as all messages with a particular label. In further embodiments, the viewport can display a custom subset of messages based on the messages that the user has explicitly selected. The selection of the viewport provides the importance prediction model 126 with messages associated with the viewport.

One user account may include two or more importance prediction models 126 to create two or more viewports for displaying or presenting sorted lists of messages. Each importance prediction model may be used to calculate separate importance scores for a particular message by using different message quality signals or by different weights for one or more message quality signals. For example, one importance prediction model may be based on a combination of a signal identifying whether the user forwards the message and a signal identifying whether the user responds to the message. The same user account may include a importance prediction model based on a combination of signals indicative of the number of times the user has read the message and signals indicative of the total time spent reading the message. When a user applies each viewport model to the same set of messages (folders, search results, labels, etc.), the two viewport models differ in the same set of messages based on importance scores calculated independently by each model. You can sort.

In some embodiments, the mechanisms for sorting the messages described below apply to conversations in conversation lists, where each conversation is a group of messages grouped together according to predefined criteria. For example, each conversation may include one or more messages that are responses to other messages in the conversation as well as the original message, and information from zero or more messages or messages in the conversation that forward messages. While some conversations may contain only one message, the list of conversations within a user account will typically be a number of conversations, each containing two or more messages. In some embodiments, the importance prediction models and viewport functions described below are applied to conversations in the list of conversations by applying these mechanisms to each message in each conversation and generating one or more combined importance scores, the importance of which Scores are used to sort conversations. Alternatively, the mechanisms described below with regard to message sorting may be in accordance with predefined criteria (eg, all messages in a conversation when the conversation includes N or less messages, otherwise the last N in the conversation). Messages, or the last message in a conversation plus N-1 longest messages, where N is a predefined value (e.g., 5)-) applies to a subset of messages in each conversation. Additional information regarding an email application that stores and displays conversation lists can be found in US Patent Application Publication 2005-0222985 A1, which is incorporated herein by reference as background information.

2 is a flow diagram of the interaction between a client and a server regarding generating and displaying messages sorted by importance scores. The selection of the viewport triggers the user request 202 to sort the messages based on the importance scores. The user request 202 provides information about the selected viewport to the importance prediction models 126 (FIG. 1) and defines a set or subset of messages in the message database 128 that the user wants to view. For example, when the selected viewport is a folder in a user account, the user request 202 includes information identifying a subset of messages belonging to this folder.

In some embodiments viewport selection may be made explicitly by the user, such as selecting a folder of messages in a messaging application for viewing. In other embodiments viewport selection may be automatically triggered by a client device that automatically selects a importance-based viewport for display at the start of a message message viewing application.

Referring again to FIG. 2, the server receives the user request and identifies a set of messages associated with the selected viewport. The server calculates (204) the importance score for each message in the selected viewport and generates an ordered message list based on the importance scores (204). The server then sends back the generated message list to the client (206), which displays the sorted list in the selected viewport (208).

In one embodiment, a second separate client device may trigger a user request 210 for a message list that is sorted by importance score. The second request is for a set of messages separate from the first request 202 but may be from the same user account. Alternatively, the second request is for exactly the same messages as the first request, but may come from the second client. In addition, the server calculates 212 the importance score according to the importance prediction model for each message in the selected viewport and generates an ordered message list based on the importance scores (212). The server may use the same importance prediction model as in 204 or may use a separate importance prediction model depending on the client viewport selected on the second client. After the server generates the ordered message list (212), the message list is sent back to the requesting second client (214). The second client then displays a list of messages generated in the selected viewport on the second client (216).

User requests may be based on manual selection of viewport by the user or may be selected based on automatically triggered viewport selection. In some embodiments, if the user is a client device such as a personal computer, the user may explicitly select folders with associated viewports or select predefined or custom viewports to be applied to any particular folder or inbox of the user account. have. In some embodiments, if the client is accessing messages in his or her user account using a PDA or cell phone, a particular viewport is automatically used for use with the device when the user attempts to view the messages. Can be selected. For example, a user may have a client application on a cell phone or mobile device that automatically selects a particular viewport (eg, a viewport that displays messages sorted according to predicted importance to the user) at startup.

Periodically, the importance prediction models for each user are updated 220 based on recent user interactions with messages through the viewports associated with the importance prediction models (ie, while using it). In one embodiment, message state database 416 stores user interactions with messages as described further below. Since each model applies a set of message quality signals to calculate importance scores for the messages, updating the models retrieves relevant interaction signals from the message state database 416 (see FIG. 4 described below). And thus applying these signals to the model. In some embodiments, this process may occur at regular time intervals, such as once every night, while in other embodiments this process may occur somewhat frequently. In some embodiments, each model may be updated at a time independent of when other models in the same user account are updated (user interaction with messages through (ie, while using) the viewport associated with each model). Based on them).

In some embodiments, importance prediction models 126 may collect message interaction signals from multiple user accounts. The models adapt over time to determine the strongest indicators of message importance for individual users. Collecting user interaction signals from multiple users provides another means of establishing importance prediction model 126. The model may update itself by retrieving the relevant interaction signals stored in the message status database 416 for multiple users and applying these signals to the model accordingly.

3 is a flow diagram of another embodiment of interactions between a client and a server regarding the generation and display of important messages in viewports. User action at the first client device triggers a user request 302 for the importance score based message list. The server receives the first request at 304, calculates importance scores for each message according to the predictive model, and generates an ordered message list. The server then sends back the generated message list to the client (306), which displays the list in the selected message viewport (308).

In one embodiment, the same client may trigger 310 a second request for a different viewport than that used in the previous step, using a separate importance prediction model. The second request may also be for a different set of messages than those requested by the first request 302. For example, the second request 310 may be for a different message folder than that selected in the first request 302.

The server receives the second request at 312, calculates importance scores for each message according to the prediction model, and generates an ordered message. However, here the server uses the importance prediction model associated with the selected client viewport. This model may be the same importance prediction model as in step 304 or the model may be different from the previous importance prediction model. In some embodiments, a user can associate the same importance prediction model with multiple folders. In such a scenario, the server will apply the same importance prediction model to the different message lists associated with the first request 302 and the second request 310.

After the server generates the current message list at 312, the server sends it back to the requesting client. The requesting client then displays this sorted message list 316 in each message viewport at the client.

In some embodiments, a user may want to apply different importance prediction models to the same set messages in a similar viewport on different clients. For example, a user may want to see important messages in a specific folder in his account while at home or at work. For example, messages that are important to the user while at work may be different from messages that are important to the user while at home. In such a scenario, selecting the same folder from similar viewports on different clients will order the messages differently by applying different importance prediction models to at least some of the same messages.

In some embodiments, the user location context information may describe the physical location of the user (eg, home, work, vehicle, etc.) and may additionally describe the current time of day. This physical location information can be defined and selected directly by the user. For example, a user can create different profiles for each location or time of day. Each profile is then associated with a different viewport, thus importance prediction models. The physical location of the client may be automatically determined using known positioning techniques, for example, the Global Positioning System (GPS). For example, the user's cellular-phone may know when the user is at home based on the user's current geolocation location and then associate the appropriate viewport and importance prediction model with the user's messages.

Referring again to FIG. 3, the second user request 310 can also provide the user location to the server. The server uses this location information to select an appropriate importance prediction model upon receipt of the second request 310. As described above, the server then continues to generate 312 a message list. By allowing the user location to help select the importance prediction model, the user can associate the same message set with multiple importance prediction models.

4 is a flow diagram of a process for generating an ordered message list of importance. At 402, a user query is received by the server for generating an ordered message list. At 404, the user query is made against the full index. The message list is then retrieved at 404 in response to a user query. The resulting message list is entered into viewport function 408 to determine individual message importance scores and produces an ordered message list at 410.

In some embodiments, the process can begin at message list 406 because the client can identify a particular set of messages about the viewport. In such case query 402 and full-text index 404 may not be required to determine message list 406 for input to viewport function 408. For example, when a user selects a viewport associated with a folder in a user account, messages for that folder may be determined using information stored in the message database 128, in which case to generate a message list 406. Full index search is not required.

In some embodiments, viewport function 408 may utilize each message in message list 406 through the use of one or more importance prediction models, contact list 412, message database 128, and message status database 416. Determine importance scores for. The message state database 416 stores state information including information indicating whether a message has been read or not. The message state database 416 can also store other state information, such as one or more of the following: the number of times each message has been opened or read, the amount of time the user has left the message open for reading, and the message The number of times you responded to, the number of times a user forwarded a message, and more. Models may be used in conjunction with contact list 412 to calculate importance scores for each message. The contact list 412 maintains affinity scores for other message participants, which will help identify important users as will be described later. Affinity scores may be used by each importance prediction model to boost the importance scores of messages sent by contacts (eg, people, companies, etc.) in the user's contact list 412. . For example, messages from contacts with high affinity scores in the user's contact list are messages from contacts with lower affinity scores and messages from people or entities not listed in the user's contact list. May receive a greater importance score boost than these. This information can be combined in the viewport function 408 with other portions of the importance score calculation performed by the importance prediction model used by the viewport function.

5A illustrates an example user account record 502 in the user account database 130 (FIG. 1) in accordance with some aspects. Some of the information that may be associated with the user account record 502 includes the message database 128 (FIG. 1), the message state database 416 (FIG. 4), the professional index 404 (FIG. 4), and the contact list ( 412 (FIG. 4). Some of this information may be stored in separate user databases, or in some embodiments may be stored in a single database for all users.

5B shows an example message record 510 in the message database 128 in accordance with some aspects. The message record 510 includes a message identifier 512 that uniquely identifies the message. In some embodiments, message identifier 512 is associated with one message. In some embodiments, message identifier 512 is associated with a group of messages that include a conversation. Some of the information that may be associated with the message 510 includes metadata 514, header information 520, and content 530.

In some embodiments, metadata 514 includes information related to the message, such as label or folder 516 assignments and time stamp 518. Each message in message database 128 may be assigned to a folder in a user account. Additionally, the user can apply one or more labels 128 to any message in the user account. Time stamp 518 provides information indicating the time at which the message was received by the user account. Timestamp 518 may be valuable for calculating the user's time spent interacting with such a message.

Header information 520 may include information identifying, for example, sender 524 and recipients 522, message reception date and time values (sometimes referred to as date values or time values) of the message, and subject 528. It may include header information of the containing message. Other information may also be included in the header information as described in RFC 2822, which is incorporated herein by reference. The message content 530 may include content of a message. Content 530 can include text, images, and attachments. Those skilled in the art will recognize other methods for storing message information. For example, the attachment may be stored in another storage structure and a reference to it may be located in message 510.

5C shows an example message status record 534 in the message status database 416 in accordance with some embodiments. Each message status record 534 includes a set of status information for a message or conversation identifier and a single message in the message database 128. In some embodiments, the status information includes one or more timestamps 536 and a set of fields representing various user interactions tracked for the representative message. One or more timestamps 536 may be used to calculate message quality signals indicative of the length of time spent by a user in performing various message interactions. For example, one timestamp may identify the first time the user read the message while another timestamp may identify the most recent time the user read the message. The first timestamp may be used to determine a delay between the receipt of the message and the user's reading of the message. The second timestamp may be used to determine the length of time between readings of the message by the user.

The set of record fields in the message status record 534 is correlated to the message quality signals used by the importance prediction models 126. For example, this set of fields may include a status as to whether a particular message has been read, a count of the number of times the message has been read, and a total time value for the user to read the message. This set of fields may also include other message quality signal indicators, such as whether the message was forwarded or responded to the message.

5D shows an example contact entry 564 in contact list 412 (FIG. 4), in accordance with some embodiments. Each contact entry 564 may represent one contact in which the user can communicate or communicate using messages as defined above. Each contact entry 564 includes information about the name of the contact 566, email contact information of the contact 568, telephone contact information 570, and one or more affinity scores 572 for the contact. One or more affinity scores 572 provide a numerical representation of user affinity for a particular contact with respect to calculating importance scores for the messages. In one embodiment, the affinity scores may affect the importance scores assigned to messages received from contacts listed in the user's contact list (eg, a person, company, or other entity). In some embodiments, each affinity score 572 represents the amount of communication activity between a contact and a user. Contributions to the affinity score (eg, communications sent to or by each contact) can be time weighted, so that more recent contributions are weighted larger than older contributions.

In some aspects, the type and perspective of the client device can affect the weight associated with the affinity score 572 because it is applied by one or more importance prediction models. For example, if a user is viewing messages on his work PDA during business hours, the affinity score 572 for contact entry 564 representing the manager is the case when the user is viewing messages on his home personal computer over the weekend. Larger weights may be assigned. Alternatively, the viewports used for these two contexts can assign different weights to each affinity score 572.

6 is a block diagram of a client device 600. In some embodiments, device 600 may include one or more processing units (CPUs) 602; One or more network or other communication interfaces 604; Memory 606; A user interface 610 (which may include a display 612 and a user input device 614) and one or more communication buses 608 for interconnecting these components. The communication buses 608 may include circuitry (sometimes called a chipset) that interconnects and controls communications between system components. Memory 606 includes fast random access memory, such as DRAM, SRAM, DDR RAM or other random access solid state memory devices, and may include one or more magnetic disk storage devices, optical disk storage devices, flash memory devices, or It may include non-volatile memory such as other non-volatile solid state storage devices. Memory 606, or alternatively non-volatile memory device (s) in memory 606, includes computer readable storage media. In some embodiments, memory 606 includes mass storage located remotely from central processing unit (s) 602. In some embodiments memory 606 also stores the following programs, modules, and data structures, or a subset thereof:

An operating system 616 including procedures for processing various basic system services and for performing hardware dependent tasks;

Other computers via one or more communication networks 104 (FIG. 1) and one or more communication network interfaces 604 (wired or wireless) such as the Internet, other wide area networks, local area networks, metropolitan area networks, and the like. A communication module 618 used to connect the client device 600 to a network;

Client application 620 for viewing and interacting with messages in viewports, including display of importance score-based message list 622; And

Other optional client applications 624.

7 is a block diagram illustrating a message server 700 according to an embodiment of the present invention. In some embodiments, message server 700 may include one or more processing units (CPUs) 702; One or more network or other communication interfaces 704; Memory 706; One or more communication buses 708 for interconnecting these components. Memory 706 includes fast random access memory, such as DRAM, SRAM, DDR RAM or other random access solid state memory devices, and may include one or more magnetic disk storage devices, optical disk storage devices, flash memory devices, or It may include non-volatile memory such as other non-volatile solid state storage devices. Memory 706, or alternatively non-volatile memory device (s) in memory 706, includes a computer readable storage medium. In some embodiments, memory 706 includes mass storage located remotely from central processing unit (s) 702. In some embodiments the memory 706 also stores the following programs, modules, and data structures, or a subset thereof:

An operating system 716 including procedures for processing various basic system services and for performing hardware dependent tasks;

Other computers via one or more communication networks 104 (FIG. 1) and one or more communication network interfaces 704 (wired or wireless), such as the Internet, other wide area networks, local area networks, metropolitan area networks, and the like. A network communication module 712 used to connect the message server 700 to a network;

User account information; A user account database 130 including a message database 130, a message state database 416, a full text index 406, and a user contact list 412;

User queries (e.g., queries that specify words or terms that may be located within the content of one or more messages and / or metadata values that may be associated with one or more messages (e.g., date, folder A search engine 122 used to provide message list results based on one or more queries) specifying a label, sender name, participant name, etc.);

A search result ranker 124 that provides a function for sorting messages based on importance scores associated with the messages;

One or more importance prediction models 126 used to generate importance scores for the messages;

Message lists 730 generated at the server for sorting purposes and for sending to clients; And

Updater module 732 for importance prediction models that process message quality signals and update the models accordingly.

Each of the elements may be stored in one or more of the previously mentioned memory devices and corresponds to a set of instructions to perform the function. The modules or programs (ie, sets of instructions) need not be implemented as separate software programs, procedures or modules, so that various subsets of these modules may be combined or otherwise re-installed in various embodiments. Can be arranged. In some embodiments, memory 706 may store a subset of the modules and data structures. The memory 706 may also store additional modules and data structures not described above.

Although FIG. 7 shows a “message server”, FIG. 7 is intended as a functional description of various features that may exist in a set of servers rather than as a structural schematic of the embodiments described herein. Indeed, and as will be appreciated by those skilled in the art, items shown separately may be combined and some items may be separated. For example, some items shown separately in FIG. 7 may be implemented on single servers and single items may be implemented by one or more servers. The actual number of servers used to implement the message server and how features are allocated between them will vary from embodiment to embodiment and depend in part on the amount of data traffic that the system must handle during peak usage periods as well as during average usage periods. You can depend on it.

For purposes of explanation, the description has been described with reference to specific embodiments. However, the illustrative discussions above are not intended to limit the invention to the precise forms disclosed or exhaustive. Many modifications and variations are possible in view of the above teachings. The embodiments have been selected and described in order to best explain the principles of the present invention and their practical applications, thereby enabling those skilled in the art to best utilize the present invention and various embodiments with various modifications suitable for the particular use envisioned. There will be.

Claims (24)

As a method for displaying messages,
In response to a first user action at a first client device, displaying a first subset of a first plurality of messages at the first client device, wherein the first subset is calculated using a first importance prediction model. Ordered by the first importance score being; And
In response to a second user action at a second client device, displaying at the second client device a second subset of a second plurality of messages, the second subset being calculated using a second importance prediction model Sorted by the second importance score being
Wherein the first and second plurality of messages comprise messages in a single user account of each user,
The first and second importance prediction models are different,
Method for displaying messages. The method of claim 1,
The first user action is a user selection of a first viewport at the first client device;
Method for displaying messages. The method of claim 1,
The first user action is a user selection of a messaging application at the first client device;
Method for displaying messages. The method of claim 1,
Calculating the first importance score for each message in the first subset of the first plurality of messages based on the first set of message quality signals; And
Calculating a second respective message importance score for each message in the second subset of the second plurality of messages based on the second set of message quality signals
And displaying the messages. The method of claim 4, wherein
The calculating operations are performed at a server remote from the first and second client devices,
Method for displaying messages. The method of claim 4, wherein
For each message, the first and second sets of message quality signals are each:
A signal identifying whether the user has read the message;
A signal identifying a delay time from receipt of said message to an initial time said user reads said message;
A signal identifying whether the user responded to the message;
A signal identifying whether the user has forwarded the message;
A signal identifying whether the user reads the message out of order;
A signal indicative of the total time taken by the user to read the message;
A signal identifying whether the user has performed a search for the message;
A signal indicating the number of times the user has read the message; And
A signal indicating the affinity of the user for another participant of the message
And one or more of the following. The method according to claim 6,
To determine membership and weights of message quality signals in the first set of message quality signals in accordance with user interactions with messages in the user account while using the viewport associated with the first importance prediction model. Updating the first importance prediction model using machine learning;
Method for displaying messages. The method of claim 7, wherein
Machine learning is used to determine membership and weights of message quality signals in the second set of message quality signals in accordance with user interactions with messages in the user account while using the viewport associated with the second importance prediction model. Updating the second importance prediction model using;
Method for displaying messages. The method according to claim 6,
A machine to determine membership and weights of message quality signals in the first set of message quality signals according to user interactions with messages in multiple user accounts while using the viewport associated with the first importance prediction model. Updating the first importance prediction model using learning,
Method for displaying messages. The method of claim 9,
To determine membership and weights of message quality signals in the second set of message quality signals according to user interactions with messages in the plurality of user accounts while using the viewport associated with the second importance prediction model. Updating the second importance prediction model using machine learning,
Method for displaying messages. The method of claim 4, wherein
The first set of message quality signals comprises at least one signal in addition to a date and time signal,
Method for displaying messages. The method of claim 4, wherein
The second set of message quality signals comprises at least one signal in addition to a date and time signal,
Method for displaying messages. The method of claim 1,
The second plurality of messages comprises at least a subset of messages in the first plurality of messages,
Method for displaying messages. The method of claim 1,
The first plurality of messages comprises messages selected from the group consisting of (A) messages in a first predefined folder of the user account, and (B) messages labeled with a first label; And
Wherein the second plurality of messages comprises messages selected from the group consisting of (A) messages in a second predefined folder of the user account, and (B) messages labeled with a second label.
Method for displaying messages. As a method for displaying messages,
On the client device:
In response to a first user action at the client device, displaying at the first client device a first subset of the first plurality of messages, the first subset being calculated using a first importance prediction model; Sorted by 1 importance score; And
In response to a second user action at the first client device, displaying a second subset of a second plurality of messages at the first client device, the second subset using a second importance prediction model; Sorted by the second importance score computed
Wherein the first and second plurality of messages each include messages in a single user account,
The first and second importance prediction models are different,
Method for displaying messages. As a method for displaying messages,
In response to a first user action at a first client device, sending information to the first client device listing a first subset of a first plurality of messages, the list of the first subset being of first importance; Aligned by the first importance score calculated using the prediction model; And
In response to a second user action at a second client device, sending information listing the second subset of the second plurality of messages to the second client device, wherein the list of the second subset is of second importance; Aligned by a second importance score calculated using the prediction model;
Wherein the first and second plurality of messages each include messages in a single user account of each user,
The first and second importance prediction models are different,
Method for displaying messages. As a method for displaying messages,
In response to a first user action at a first client device, sending information to the first client device listing a first subset of a first plurality of messages, the list of the first subset being of first importance; Aligned by the first importance score calculated using the prediction model; And
In response to a second user action at the first client device, sending information to the first client device listing a second subset of a second plurality of messages, the second subset predicting a second importance Aligned by a second importance score calculated using the model;
Wherein the first and second plurality of messages each include messages in a single user account of each user,
The first and second importance prediction models are different,
Method for displaying messages. A computer readable storage medium for storing one or more programs, the method comprising:
The one or more programs are:
At the first client device,
In response to a first user action at the first client device, display a first subset of a first plurality of messages at the first client device, the first subset computed using a first importance prediction model Ordered by the first importance score being; And
In response to a second user action at the first client device, for displaying at the first client device a second subset of a second plurality of messages, the second subset using a second importance prediction model; Sorted by the second importance score computed
Instructions, wherein the first and second plurality of messages each include messages in a single user account,
The first and second importance prediction models are different,
A computer readable storage medium storing one or more programs. A computer readable storage medium for storing one or more programs, the method comprising:
The one or more programs are:
In response to a first user action at a first client device, send information to the first client device listing a first subset of a first plurality of messages, the list of the first subset predicting a first importance Aligned by the first importance score calculated using the model; And
In response to a second user action at a second client device, information for transmitting a second subset of the second plurality of messages to the second client device, wherein the list of the second subset is of second importance. Aligned by a second importance score calculated using the prediction model;
Instructions, wherein the first and second plurality of messages each include messages in a single user account of each user,
The first and second importance prediction models are different,
A computer readable storage medium storing one or more programs. A computer readable storage medium for storing one or more programs, the method comprising:
The one or more programs are:
In response to a first user action at a first client device, send information to the first client device listing a first subset of a first plurality of messages, the list of the first subset predicting a first importance Aligned by the first importance score calculated using the model; And
In response to a second user action at the first client device, information for transmitting a second subset of the second plurality of messages to the first client device, the second subset predicting a second importance Aligned by a second importance score calculated using the model;
Instructions, wherein the first and second plurality of messages each include messages in a single user account of each user,
The first and second importance prediction models are different,
A computer readable storage medium storing one or more programs. A system for displaying messages,
Memory;
One or more processors; And
One or more modules stored in the memory and configured to be executed by the one or more processors
The module includes:
In response to a first user action at a first client device, display a first subset of a first plurality of messages at the first client device, wherein the first subset is computed using a first importance prediction model; Sorted by first importance score; And
In response to a second user action at a second client device, for displaying at the second client device a second subset of a second plurality of messages, the second subset computed using a second importance prediction model Sorted by the second importance score being
Instructions, wherein the first and second plurality of messages each include messages in a single user account of each user,
The first and second importance prediction models are different,
System for displaying messages. As a client device,
Memory;
One or more processors; And
One or more modules stored in the memory and configured to be executed by the one or more processors
Wherein the one or more modules are:
In response to a first user action at the client device, instructions for displaying at the client device a first subset of a first plurality of messages, the first subset being calculated using a first importance prediction model. Sorted by first importance score; And
In response to a second user action at the client device, instructions for displaying a second subset of a second plurality of messages at the client device, wherein the second subset is calculated using a second importance prediction model. Sorted by second importance score ―
Wherein the first and second plurality of messages each include messages in a single user account,
The first and second importance prediction models are different,
Client device. As a server system,
Memory;
One or more processors; And
One or more modules stored in the memory and configured to be executed by the one or more processors
Wherein the one or more modules are:
In response to a first user action at a first client device, instructions for sending to the first client device information listing the first subset of the first plurality of messages, the list of the first subset being selected from the first user device. Ordered by the first importance score calculated using the 1 importance prediction model; And
In response to a second user action at a second client device, instructions for sending to the second client device information listing a second subset of a second plurality of messages, wherein the list of second subsets is determined by the second user device. Aligned by a second importance score calculated using a two importance prediction model;
Wherein the first and second plurality of messages each include messages in a single user account of each user,
The first and second importance prediction models are different,
Server system. As a server system,
Memory;
One or more processors; And
One or more modules stored in the memory and configured to be executed by the one or more processors
Wherein the one or more modules are:
Instructions, in response to a first user action at a client device, to send information to the client device listing a first subset of a first plurality of messages, the list of the first subset being a first importance prediction model. Aligned by a first importance score computed using; And
Instructions, in response to a second user action at the client device, to send information to the client device listing a second subset of a second plurality of messages, the second subset generating a second importance prediction model. Aligned by a second importance score computed using;
Wherein the first and second plurality of messages each include messages in a single user account of each user,
The first and second importance prediction models are different,
Server system.

Images